video: Researchers at Colorado State University have developed a flexible robot that functions like an elephant’s trunk to grip objects, with a goal of one day sending teams of these machines to aid in space exploration or to help with any number of other complex tasks.
Credit: Colorado State University Walter Scott, Jr. College of Engineering
EMBARGO: This content is under embargo until Nov. 11 at 5 a.m. (U.S. Eastern Time). Reporters may conduct interviews and request copies of the paper or multimedia assets in advance of that date. But they may not publish or share this information on any public platform before the embargo lifts.
Researchers at Colorado State University have developed a flexible robot that functions like an elephant’s trunk to grip objects, with a goal of one day sending teams of these machines to aid in space exploration or to help with any number of other complex tasks.
The robots are described in detail in Nature Machine Intelligence and are designed to address limitations in current robotic systems, which are more commonly built from rigid blocks that limit the shapes they can form.
These new robots were made instead with a design approach called tensegrity, in which stiff rods are held in place by a web of tensioned cables rather than being rigidly locked together. This makes them flexible and lets them bend, collapse and change shape as needed. Like tentacles, they have no traditional joints yet feature a pair of gripping claws at both ends of their tube-like bodies that can attach to partner robots for teaming or to manipulate items.
The robots were developed by Associate Professor Jianguo Zhao and his team in the Department of Mechanical Engineering at CSU. He described the robots as generalists that could one day help with search and rescue efforts or support humans in a manufacturing setting, among other examples.
“Often the physical structure of robots is fixed, but our approach allows these flexible systems to adapt their shape to their environment and to the potentially changing needs of their mission by collapsing, stretching or grasping,” he said. “They are also modular, meaning each robot works like a building block: It can operate on its own or connect with others to form bigger, more capable shapes.”
He added that the dual claw-attachment design avoided the need for magnets, which can sometimes fail as connection points. Zhao said the combination of claw attachments and body structure open many possibilities for their eventual use.
“They can bend around to lift and carry different sized objects with their body or to navigate difficult terrain like a snake would,” Zhao said. “The claw-attachment also reduces the chance of misalignments during connections – making it easier for them to team up and form loops, chains and branches of different sizes to accomplish their tasks. They are very adaptable to the situation and goals.”
Mahmud Hasan Saikot is a doctoral student at CSU and co-first author on the paper. He took over the robot initially designed by team member Sydney Spiegel, the paper’s other co-first author. Saikot said the team went through several design ideas before landing on this version of the robotic arms. One of the challenges was finding motors that were small enough to support the various types of movements but were also powerful and reliable over longer periods of time. He said the team spent a lot of time testing configurations and taking the robots into the field to see how they handled crossing natural landscapes in addition to work in the lab.
The next step in the project, he said, will be developing the ability for these systems to work with humans and on their own. The team also hopes to make them operate at faster speeds.
“Right now, these systems need to be piloted by a human. However, we have already demonstrated that two robots can find each other and connect on their own. In the future, we can envision mostly autonomous teams of tens or even hundreds of them – attached through the claws – traveling to a location and then deploying for a specific purpose,” he said. “With that in mind, we want to continue improving their ability to work together with minimal human supervision.”
Other CSU-affiliated researchers listed as authors on the paper include Sudheera Akalanka Kariyawasam, Andrew Stefka and Josh Chrisler.
Journal
Nature Machine Intelligence
Article Title
Tensegrity continuum robots enable task-adaptive morphologies for cooperative behaviours
Article Publication Date
11-Nov-2026